Int. J. Plant Sci. 160(5):887–896. 1999. q 1999 by The University of Chicago. All rights reserved. 1058-5893/1999/16005-0010$03.00

FECUNDITY ABOVE THE SPECIES LEVEL: OVULE NUMBER AND BROOD SIZE IN THE GENISTEAE (FABACEAE: PAPILIONOIDEAE) Javier Herrera1 Departamento de Biologı´a Vegetal y Ecologı´a, Universidad de Sevilla, Apartado 1095, E-41080, Seville, Spain

Per-fruit components of fecundity (ovules per ovary, seeds per fruit, seed : ovule ratio, and patterns of seed maturation within pods) were studied in 33 species of the Genisteae, a legume tribe of mostly Mediterranean shrubs, e.g., Cytisus, Ulex, that also includes more widespread herbs such as Lupinus. Species identity explained most variance in both potential (ovules) and realized (seeds per fruit and seed : ovule ratio) fecundity, and although significant population-specific effects existed, these were relatively small, indicating that estimates of fecundity were not particularly prone to environment-induced changes. Average seed : ovule ratios varied between 20% (for a few species in which fixed rates of embryo abortion result into single seeded pods) and 100% (annuals), but most taxa were in the 40%–60% range. The probability that an ovule near the base of the ovary matured into a seed was significantly lower than the pod average in the majority of species. As indicated by the large size of stigmatic pollen loads, this was unlikely to result from insufficient pollination but rather from preferential embryo abortion. High abortion rates next to pod base occurred in both presumed selfers and obligate outcrossers. Evidence points to relatively stereotyped reproductive ways and high reproductive coherence in the tribe, with major departures from the norm being high seed : ovule ratios exhibited by annual Lupinus (associated with selfing), and low ratios shown by species with single seeded pods (associated with fixed rates of embryo abortion). Keywords: Mediterranean, shrublands, seed-set, abortion, reproduction.

Introduction

affects this kind of attribute is weaker or stronger than overall trait covariation. In other words, there is no knowledge of the context of phylogenetic constraint affecting other plant traits. To know how and whether common phylogeny is posing a severe constraint on the evolution of reproductive attributes, it could be useful to see how neat is the group pattern exhibited by ecologically determined characteristics relative to the group behavior regarding other traits. Suppose, for example, that we find a very distinct pattern at the family level for a given characteristic (e.g., number of ovules in the ovary) that becomes blurred or nonexistent for its environmentally determined aftermath (seed number). If this happens, the importance of phylogenetic constraints should be considered low with regard to ecological forces. Conversely, a similar pattern for both sets of traits would indicate that phylogenetic cohesion is so strong that it can indeed override environmental effects. Biosystematists have argued that some plant organs can be more evolutionarily constrained than others. In the Leguminosae, for example, there are more morphological and functional modifications in fruits than in flowers or seeds (Polhill et al. 1981), a situation reflected in the diversification of this family. In this article, I report on how unrestrained are several components of fecundity (ovule number, seed number, seed : ovule ratios, and seed abortion patterns) that may be subject to ecological effects to varying degrees. The study group is the tribe Genisteae (Fabaceae: Papilionoideae), a lineage with about 500 species of which ca. one-half belong to 18 genera of mainly Mediterranean shrubs (the so-called GenistaCytisus complex), while the remaining are annual and perennial herbs in the genus Lupinus (America and Europe). Di-

Researchers have sometimes found evidence that a number of reproductive attributes of plants can be subject to considerable phylogenetic constraint (Herrera 1989; Eriksson and Bremer 1991, 1992; but Lord et al. 1995). Investigations on character correlation or syndromes in plants have also provided support for this view (Herrera 1992; Fischer and Chapman 1993; Jordano 1995). Relatively invariant reproductive patterns in a plant group have the potential to limit the development of coevolutionary relationships with mutualists, which makes the issue of phylogenetic constraint relevant to the debate on whether pollination and seed dispersal systems should in general be regarded as specialized or generalized (Herrera 1995; Waser et al. 1996; Waser 1998). Most studies on group-specific reproductive patterns have focused on traits directly involved in mutualistic relationships such as flowering and fruiting phenology (Kochmer and Handel 1986; Smith-Ramirez et al. 1998) and life-form and breeding system (Renner and Ricklefs 1995; Buide et al. 1998). Fruit attributes that have an obvious bearing on seed dispersal have also been viewed in this light (e.g., Jordano 1995). Seemingly, however, morphogenetic constraints should affect all plant attributes, not just those regulating relationships with other organisms or the environment. The problem of focusing only on such traits when studying phylogenetic constraints is that you cannot know whether the correlation-covariance pattern that 1

E-mail [email protected].

Manuscript received January 1999; revised manuscript received May 1999.

887

INTERNATIONAL JOURNAL OF PLANT SCIENCES

888

agnostic features of this tribe include a bilabiate calyx and monadelphous stamens with dimorphic anthers (Bisby 1981). The following questions are addressed: Is there a group pattern for potential (ovules) and realized (seeds) fecundity? How variable are pod-filling patterns among taxa of this group? Do breeding systems and filling patterns correlate in some way? Answers to the latter question may contribute to the current knowledge on male gametophyte selection mechanisms and maternal screening of progeny quality in the Leguminosae (Bawa and Buckley 1989).

Study Area Research was carried out across the western half of Andalucia (southern Spain), a region of ca. 45,000 km2 that includes the two mountain ranges bordering the lower Guadalquivir River Valley and an extensive coastal area near the Atlantic Ocean. The region has a typical Mediterranean climate with dry summers and mild, rainy winters and supports a variety of sclerophyllous scrub types on sand dunes, riversides, as well as limestone and sandstone mountains. Maximum and minimum distances among any two study localities were 210 km and 19 km. Their names and the corresponding letter codes used are listed in a note to table 1.

Study Organisms Most Genisteae in the study area are shrubs, often spiny, which typically develop after destruction or disturbance of evergreen-oak forests. They are sometimes dominant in scrub, e.g., species of Ulex, but may also represent just one component of diverse, mixed formations including taxa of several

families. Genisteae have good sprouting abilities following fire and disturbances in general, which contrasts with other cooccurring, frequent scrub components, e.g., the Cistaceae. Over the geographical scale of this study, richness of Genisteae at a given site increased with increasing soil fertility and rainfall (Arroyo and Maran˜o´n 1990). Most studied taxa bloom in spring, March–May (Herrera 1986; Arroyo 1990), and disperse seeds during the summer. Less frequently, autumn and winter flowering also occurs, e.g., in Ulex species. The mostly yellow flowers are chiefly pollinated by bees and, with few exceptions (e.g., Retama species), produce no nectar (Herrera 1985). Experimental hand pollinations have shown that conventional self-incompatibility is absent but, at least in shrubby taxa, some form of postzygotic barrier to inbreeding seems to exist (Rodriguez Rian˜o 1997; P. E. Gibbs, unpublished data). This, along with the vast numbers of flowers produced by individual plants, hundreds to thousands, results in extensive fruit abortion at early stages. Insect exclusion from flowers customarily reduces fruit production to values near zero (e.g., 0.2% in Stauracanthus genistoides, 0% in Ulex minor, and 0% in Cytisus grandiflorus [Herrera 1987]; 0% in Retama sphaerocarpa and 0% in Cytisus striatus [Rodriguez Rian˜o 1997]). Annual species of Lupinus, however, may set abundant fruit even inside an insectproof greenhouse (S. Talavera, personal communication). Pod size in the studied group ranges from 5 mm long in Genista triacanthos to 80 mm in Spartium junceum. At least in some genera (Cytisus, Ulex, Stauracanthus), carunculate seeds are released violently from pods and may then be gathered by foraging ants (Pijl 1972; Bossard 1991; J. Herrera, personal observation).

Table 1 Between-Population Variations for the Number of Ovules in the Ovary, Seeds per Fruit, and Seed Set in Genisteae Species

Localitiesa

Adenocarpus telonensis ... ... .. Calicotome villosa .............. Cytisus baeticus .. .. ............. Cytisus grandiflorus ............ Genista cinerea . ................ Genista polyanthos ............. Genista triacanthos ............. Genista umbellata . ............. Retama monosperma .... ... ... . Spartium junceum . ... .... ... ... Stauracanthus genistoides ...... Teline liniifolia .. ... ... .... ... ... Teline mompessulana ... ... ... .. Ulex argenteus .. ... ... .... ... ... Ulex australis . . .. . .............. Ulex minor . .. . .. . .. .............

A B (150) D E (52) D F (52) G A (56) K L (120) N O (85) A B (150) P Q (79) T G (33) K U (95) T R (38) B E (51) B E (51) V W (20) X R (19) Y Z (22)

Ovules/ovaryb 7.0 9.0 12.0 13.4 6.0 7.3 4.0 4.9 5.8 15.5 5.4 4.5 7.8 5.0 5.4 6.0

(0.1) (0.1) (0.2) (0.2) (0.1) (0.2) (0.1) (0.1) (0.1) (0.2) (0.1) (0.1) (0.1) (0.1) (0.1) (0.1)

6.7 8.9 12.4 11.8 5.7 5.9 4.9 4.5 5.9 18.3 4.1 3.7 7.8 6.2 6.3 6.4

(0.1) (0.1) (0.2) (0.1) (0.1) (0.1) (0.1) (0.1) (0.1) (0.2) (0.1) (0.1) (0.1) (0.1) (0.1) (0.1)

Seeds/fruitb ns ns ns

∗

ns

∗

∗∗

ns ns

∗∗ ∗∗ ∗∗

ns

∗ ∗

ns

3.5 3.8 6.2 6.2 2.3 2.5 1.0 2.5 1.1 12.8 2.7 2.5 4.1 2.5 2.3 2.1

(0.2) (0.3) (0.3) (0.5) (0.1) (0.3) (0) (0.1) (0) (0.3) (0.2) (0.1) (0.2) (0.1) (0.1) (0.1)

4.3 (0.2) 3.8 (0.3) 5.2 (0.4) 3.4 (0.3) 1.8 (0.1) 1.6 (0.1) 1.0 (0) 2.0 (0.1) 1.0 (0) 12.9 (0.5) 2.0 (0.1) 1.6 (0.1) 2.9 (0.2) 2.1 (0.1) 2.7(0.1) 2.5 (0.2)

% Seed setb ns ns ns

∗∗

ns ns ns ∗

ns ns

∗

∗∗

ns ns ns ns

49 42 43 47 40 33 20 52 20 83 51 57 53 51 43 35

(3) (3) (3) (3) (2) (3) (1) (3) (1) (2) (3) (3) (3) (3) (3) (2)

66 43 52 29 32 29 16 43 17 71 49 44 38 34 43 39

(3) (3) (3) (2) (3) (2) (1) (3) (0) (3) (3) (2) (3) (2) (2) (3)

∗

ns ns

∗

ns ns ns ns ns ns ns ns ns ns ns ns

Note. Values are plant means (s.e.m.) based on five individuals per population. Significance for population means differences are also indicated (Student’s t-test). a The letter codes corresponding to location are A, Valverde del Camino; B, Puerto Galis; C, Carcabuey; D, Ubrique; E, Los Barrios; F, El Coronil; G, El Rompido; H, Estepona; I, Cabra; J, Aracena; K, Los Villares; L, Atajate; M, Oje´n; N, Tharsis; O, El Garrobo; P, Rute; Q, Coı´n; R, Hinojos; S, Olivares; T, Mazago´n; U, El Burgo; V, Cartaya; W, Punta Umbrı´a; X, Aznalca´zar; Y, Almonte; Z, Asperillo; MT, Montoro. Numbers in parentheses are distances in km. b ns = not significant. Numbers in parentheses are s.e.m. ∗ P ! 0.05. ∗∗ P ! 0.01.

HERRERA—BROOD SIZE IN THE GENISTEAE Table 2 MANOVA on Components of Fecundity for the 16 Genisteae Species in Table 1 Univariate statistics r2

Source

SS

df

F

Variable: Ovules/ovary .... .. 0.958 Species 4.6 15 189.1 Population 0.2 16 5.9 Error 0.2 128 Seeds/fruit . .. .. .... 0.914 Species 11.9 15 86.2 Population 0.6 16 4.2 Error 1.2 128 Seed set . .. . .. .. .... 0.774 Species 11,322.5 15 25.72 Population 1572.5 16 3.35 Error 3755.9 128

889

are linearly arranged inside the ovary, so that the positions (rank order relative to the style) where seeds had formed were easily noted. Ovular positions were then individually tested for significant departures from average seed set by comparing the frequency observed at that position against 95% confidence intervals of the species’ mean. In general, I censused 10 fruits per plant and a total of 50 fruits per population, although in several taxa with single seeded fruits (R. sphaerocarpa, Genista haenseleri, and Genista triacanthos) I examined a larger number in order to increase the sample size of low-frequency ovular positions and to compensate for empty fruits, i.e., pods that were externally normal but contained no viable seed. These were in general rare, as were those attacked by insect predators. Both types of pods have been excluded from the analysis.

Multivariate statistics Wilk’s l Source: Species ... ... .... ... Population ... .....

0.004 0.331

F 43.6 3.5

df 45/375 48/375

Note. Factor population is nested under “species” in this analysis. For all F statistics shown, the significance level is !0.001.

Methods Fecundity Variations in per-fruit components of fecundity were studied in 33 species representing some 70% of the Genisteae in the area (nomenclature and species numbers follow Tutin et al. 1980; see Talavera 1999 for a more recent taxonomic account). Of these, 17 were studied in two populations and the remaining 16 only in one. Collected fruits were always near maturity and increases in size or abortion were not expected thereafter, a stage that was established unambiguously by visiting each site at least twice. No attempt was made to obtain estimates of fruit : flower ratios (i.e., fruit set) since this would have required repeated visits to census plants. Given the distances between localities, the relatively synchronous flowering, and the number of species involved, this would have been impracticable. Moreover, fruit : flower ratios seemed to be rather low as a whole (for open pollinated flowers in natural populations: 12% in Cytisus grandiflorus, 5% in Ulex parviflorus, and 16% in Ulex minor on a single year [Herrera 1987]; 5% and 6% in Retama sphaerocarpa and 13% and 20% in Cytisus striatus, in two consecutive years [Rodriguez Rian˜o 1997]) and prone to substantial yearly variations (Rodriguez Rian˜o 1997). As a result, perflower estimates of fecundity for a few individuals per population in a single year would probably not have contributed much to establish general fertility patterns. Instead, per-fruit components, which were expected to be less subject to individual and local variations, were used throughout. Except for the rare Cytisus malacitanus, of which only four plants could be located, fruits were examined from at least five randomly chosen individuals per population. Fruits were collected at random from several branches per plant and kept in separate bags until ovules and seeds were scored. Ovules

Stigmatic Pollen Loads Pollination intensity was studied in five species, including two with few-seeded fruits (Retama monosperma and Ulex eriocladus), two with an intermediate (modal for the Genisteae) number of seeds per fruit (Stauracanthus genistoides and Genista polyanthos), and one with many-seeded fruits (Genista falcata). In all of these species, and after dissecting fruits to count seeds and ovules, I separated the still attached style, softened it with NaOH for a few minutes, and placed it on a slide with a drop of aniline blue. Pollen grains adhered to the stigma were then counted under UV light with a fluorescence microscope. Numbers reported represent minimum estimates, since an unknown fraction of the grains originally deposited on the stigma might have detached by the time the fruits were collected.

Data Analysis In the analysis of among-population variability variates are plant means. Before MANOVA (PROC GLM in SAS 1990), these were either log (ovules and seeds per fruit) or arcsin (seed : ovule ratios) transformed. Further analyses of variations among taxa used fruit means as variates and, if data from 11 population per species were available, these were combined into a single sample.

Results The studied species set abundant fruit except for Genista hirsuta. At population O of this species, I could gather only 66 developed pods from five plants and, of these, 44 contained aborted seeds only. At population MT (160 km away) I could collect 107 fruits from five shrubs, but 90 contained only aborted seeds. Discarding empty pods, in population O there were one seed and five undeveloped ovules per fruit on average, and the seed : ovule ratio was 17% (n = 22). Averages for population MT were one seed and six undeveloped ovules per fruit, and a seed : ovule ratio of 15% (n = 17). I judged both viable seed pod samples insufficient, and this species was omitted from the analyses below.

INTERNATIONAL JOURNAL OF PLANT SCIENCES

890

Table 3 Numbers of Ovules per Ovary, Seeds per Fruit, and Percent Seed Set in 32 Species of Genisteae Species Adenocarpus telonensis ......... .. Argyrolobium zanonii ... ... ... .... Calicotome villosa ... .... ... ... ... . Chamaespartium tridentatus . ... .. Cytisus baeticus .. ... ... .... ... ... .. Cytisus fontanesii .. ... ... .... ... ... Cytisus grandiflorus .. ... .... ... ... Cytisus malacitanus . ... .... ... ... .. Cytisus reverchonii . ... .... ... ... .. Cytisus striatus ... ... ... .... ... ... .. Cytisus villosus .. ... ... ... .... ... ... Genista cinerea ... ... ... .... ... ... .. Genista falcata .. ... ... .... ... ... ... Genista haenseleri ... .... ... ... ... . Genista polyanthos ................ Genista triacanthos ................ Genista umbellata . ................ Lupinus albus .. ... ................. Lupinus angustifolius ... .... ... ... . Lupinus hispanicus .. .... ... ... ... . Lupinus luteus ... ... ... .... ... ... .. Retama monosperma . .... ... ... ... Retama sphaerocarpa ............. Spartium junceum . ................ Stauracanthus boivinii .. ... ... ... .. Stauracanthus genistoides . ... ... .. Teline liniifolia ... ... ... ... ... ... ... Teline mompessulana .. ... ... ... ... Ulex argenteus . ... ... ... ... ... ... .. Ulex australis .. ... ... ... ... ... ... ... Ulex eriocladus ... ... ... ... ... ... ... Ulex minor ... ... ... ... ... ... ... ... . Means for all taxa . ... ... ... ... ... .

Ovules/ovary 6.9 4.7 9.0 5.3 12.2 7.0 12.6 7.2 11.5 9.2 10.0 5.9 22.8 8.3 6.6 4.4 4.7 5.5 6.1 4.4 4.3 5.9 4.7 16.9 2.5 4.7 4.0 7.8 5.6 5.9 7.1 6.2 7.5

(0.1–11) (0.1–18) (0.1–10) (0.1–15) (0.1–10) (0.1–14) (0.1–11) (0.1–11) (0.1–8) (0.1–9) (0.2–11) (0.1–12) (0.3–11) (0.1–12) (0.1–20) (0.1–18) (0.1–13) (0.1–10) (0.1–9) (0.1–11) (0.1–12) (0.1–15) (0.1–14) (0.2–12) (0.1–24) (0.1–20) (0.1–17) (0.1–10) (0.1–18) (0.1–15) (0.1–8) (0.1–14) (0.7–55)

Seeds/fruit 3.9 4.1 3.8 2.3 5.7 1.9 4.8 2.2 3.2 4.0 5.4 2.1 9.5 1.0 2.1 1.0 2.3 4.6 5.0 4.0 3.9 1.0 1.1 12.9 1.6 2.4 2.0 3.5 2.3 2.5 1.7 2.3 3.4

Seed set

(0.2–39) (0.1–26) (0.2–53) (0.2–49) (0.3–43) (0.1–45) (0.3–62) (0.2–70) (0.3–57) (0.2–43) (0.3–34) (0.1–48) (0.5–33) (0–0)a (0.2–71) (0–0)a (0.1–43) (0.1–23) (0.2–24) (0.1–18) (0.1–15) (0.02–23)a (0–26)a (0.3–24) (0.1–45) (0.1–47) (0.1–49) (0.2–48) (0.1–43) (0.1–39) (0.1–58) (0.1–45) (0.4–72)

58 86 42 44 47 27 38 31 28 45 53 36 42 12 31 23 48 85 82 91 91 18 23 77 63 50 49 45 42 43 25 37 47

(2–40) (2–20)b (2–53) (3–52) (2–44) (2–43) (2–58) (3–73) (2–58) (3–42) (2–32) (2–50) (2–33) (0–11) (2–63) (1–20) (2–41) (3–23)b (3–22)b (2–15)b (2–13)b (1–38) (1–31) (2–24)b (4–42) (2–42) (2–46) (2–49) (2–48) (2–41) (2–58) (2–42) (4–47)

n 100 50 100 50 100 50 100 45 50 50 50 100 50 100 100 168 100 59 50 60 50 120 100 100 50 100 120 100 98 100 50 100

Note. Values are fruit means (s.e.m. 2 coefficient of variation). n = total number of fruits examined. a Values correspond to species with largely single seeded legumes. b Seed to ovule ratios that are unusually high (and indicative of selfing).

Variations among Populations

Variations among Species

The numbers of ovules per ovary differed significantly among populations in eight species, but only four species showed significant variations in the number of seeds per fruit (table 1). Average seed : ovule ratios differed between conspecific populations in two species. However, population-specific variability seemed unrelated to spatial distance: e.g., conspecific plants 150 km apart might have either similar (e.g., Adenocarpus telonensis) or significantly different (e.g., Genista triacanthos) mean ovule numbers, and the same applied to populations separated 20 km or less (e.g., Ulex species). A more detailed analysis (MANOVA; table 2) of between population variations, incorporating jointly all three studied fecundity components as predicted by species and population factors, indicates that species accounts for most of the overall variance, but the effect of population is still significant. The variance expressed at the population level is modest (respectively, 5%, 7%, and 13%, for ovules, seeds, and seed set; PROC VARCOMP in SAS).

Mean ovule numbers (table 3) were three to 23, but most species had seven or fewer, and the frequency distribution is markedly skewed (fig. 1). In contrast, seed numbers are much more concentrated around the modal value of three and, with only two exceptions, all species had seven or fewer. Seed number was a fixed trait in two Genista species and the two species of Retama, all with single seeded fruits (although the number of ovules was far greater than one in such taxa; table 3). Yet another instance of monospermic fruits developing from several-ovuled ovaries was the abnormally infertile G. hirsuta (see above). Seed sets ranged from 10% to 100%, but there was a distinct modal value ca. 50%, which is coincident with the median of the frequency distribution (fig. 1, bottom). The lowest values were found in species with single seeded fruits and the largest ones in annual Lupinus (table 3). Species with single seeded fruits tended to have both relatively low (ca. 20%) and constant (CV ca. 25%) seed sets. (Because the number of ovules

HERRERA—BROOD SIZE IN THE GENISTEAE

891

often, the basal half of the ovary showed a lower probability of setting seeds, while the distal (stylar) half either showed a higher-than-average probability (11 species) or no significant deviation from the mean. Usually, the probability that the basal ovule would set a seed was well below the lower 95% confidence limit of average probability (fig. 3; note that these are just examples of the more usual pattern, chosen to illustrate the fact that the trend exists regardless of ovule number, seeds per fruit, seed set, growth form, or breeding system). In addition, low probability of maturing into seed often was not restricted to the basal ovule but extended to subsequent ovules, too, with up to four rank values in Spartium junceum and two rank values in G. triacanthos (note that the former has many-seeded legumes, while in the latter these are invariably one-seeded). It is also noteworthy (fig. 3) that the filling profile of Lupinus albus (a presumably selfing annual) is similar to that of Cytisus baeticus (an outcrossing shrub). There were few exceptions to the dominant pattern (fig. 4). In some of these, the trend to decreased seed set near the base of the pod still existed (A. telonensis, Retama monosperma), but it failed to reach statistical significance. More truly exceptional were the profiles of Lupinus hispanicus and Lupinus angustifolius, in which ovules closest to style had the lowest probability of resulting into a seed.

Pollination Intensity

Fig. 1 Frequency distributions for components of fecundity considered in this article.

in the ovary is not fixed, some variation exists in spite of seed number being constant.) Species in which seed number is variable, however, show a distinct pattern of decreasing variability with increasing success in ovule-to-seed transformation; i.e., the larger the proportion of ovules that develop into seeds, the smaller the variation from one pod to another (Spearman rank correlation R = 20.514, P ! 0.01 , n = 28 ). This is likely to be, at least in part, the outcome of breeding systems that profit from self-fertilization to varying degrees: four annuals and two presumed perennial-selfers stand out with very high and constant seed sets, while the remaining taxa make up a continuum of success versus variability values (fig. 2).

Estimates of stigmatic pollen loads for five taxa (table 5) show that, in spite of stigmas being relatively small, they commonly have many pollen grains attached (often 1100 in Genista falcata) and, invariably, in excess of the number of ovules. The ratio of pollen grains to seeds reached a maximum in the single seeded pods of R. monosperma. The number of pollen grains on the stigma correlated significantly with the number of seeds eventually produced by the fruit only in Genista polyanthos (R = 0.295, P ! 0.01, n = 87).

Variations within the Ovary In most species, certain ovule positions were more likely to produce mature seeds than others, and some intraovary asymmetry (i.e., at least one ovular position had a probability below or above the mean) was found in 24 species (72%). When the probability that an ovule set a seed was tested for independence against the distance (rank order) to the style, significant heterogeneity was detected in 21 (65%) of species (table 4). Most

Fig. 2 The relationship between average seed sets and their corresponding coefficients of variations in 32 Genisteae species. Open squares and triangles designate taxa with monospermic fruits and presumed selfers, respectively.

892

INTERNATIONAL JOURNAL OF PLANT SCIENCES Table 4 Within-Ovary Patterns of Seed Production in Genisteae Probability of transforming an ovule into seed relative to averagea

Position # probability independence

Adenocarpus telonensis .. ... ...... Argyrolobium zanonii ............. Calicotome villosa . ... ... ... ....... Chamaespartium tridentatus ...... Cytisus baeticus .................... Cytisus fontanesii .................. Cytisus grandiflorus ............... Cytisus malacitanus ................ Cytisus reverchonii ................ Cytisus striatus ..................... Cytisus villosus ..................... Genista cinerea ..................... Genista falcata ..................... Genista haenseleri ................. Genista polyanthos ................ Genista triacanthos ................ Genista umbellata ................. Lupinus albus ...................... Lupinus angustifolius .............. Lupinus hispanicus ................ Lupinus luteus ..................... Retama monosperma .............. Retama sphaerocarpa ............. Spartium junceum ................. Stauracanthus boivinii ............. Stauracanthus genistoides ......... Teline liniifolia ..................... Teline mompessulana .............. Ulex argenteus ..................... Ulex australis ....................... Ulex eriocladus ..................... Ulex minor .........................

x2

df

P

Stylar half

Basal half

6.4 4.1 9.6 5.7 43.1 8.6 16.0 2.7 38.4 3.4 22.0 43.8 37.3 39.0 18.5 34.6 13.0 37.1 48.2 13.1 9.9 7.0 11.3 109.4 2.0 23.9 15.0 38.8 12.8 14.6 10.3 30.6

6 4 9 5 12 6 12 6 11 8 9 5 23 8 6 4 4 5 5 4 3 5 4 18 2 4 3 7 5 5 6 6

ns ns ns ns

ns ns ns ns Higher ns ns ns Higher ns ns Higher Lower Higher ns Higher ns Higher High/Low Lower ns ns ns Higher ns Higher Higher Higher ns ns ns Higher

ns ns Lower Lower Lower ns Lower ns Lower ns Lower Lower Lower Lower Lower Lower Lower Lower High/Low ns Higher ns Lower Lower ns Lower Lower Lower Lower Lower ns Lower

∗∗∗

ns ns ns

∗∗∗

ns

∗∗

∗∗∗ ∗ ∗∗∗ ∗∗ ∗∗∗ ∗∗ ∗∗∗ ∗∗∗ ∗ ∗

ns

∗

∗∗∗

ns

∗∗∗ ∗∗ ∗∗∗ ∗ ∗∗

ns

∗∗∗

Note. For each taxon, the positions of ovules in the ovary were tested for independence against the probability of resulting into a seed. Significance of the test indicates nonevenness in fecundity. The direction of the asymmetry (either lower or higher than average) is also indicated. ns = not significant. a “Higher” and “lower” mean that at least one ovular position presented a rate of seed production respectively above or below 95% confidence intervals of the mean. In a few instances (Calicotome villosa, Chamaespartium tridentatus, Cytisus grandiflorus), there were positions in which the probability of setting seed was significantly lower than average, but the trend was not strong enough for the position # probability independence test to detect any significant effect. ∗ P ! 0.05. ∗∗ P ! 0.01. ∗∗∗ P ! 0.001.

Discussion Although related, ovule number, seeds per fruit, and seed set offer different quantitative views of the female component of plant reproduction (reviews in Lloyd 1980; Stephenson 1981; Stephenson and Bertin 1983; Primack 1987; Lee 1988; Burd 1994). Ovule number reflects a gamete packaging decision that may have consequences on the way sibling embryos are nourished and interact with each other. The seed complement of fruits, however, is determined not only by commitments derived from dispersal, development, and fertilization but also from environment-dependent factors such as polli-

nator and resource availability that are beyond direct plant control.

Population Differences in Ovule and Seed Number If realized fecundity is more environmentally determined than potential fecundity (Burd 1994, 1995), one would expect estimates of seeds per fruit and seed set to be more variable among conspecific populations than ovule number. In the studied Genisteae this is supported by data, with variance expressed at the population level increasing from ovule number (5%), to seed number (7%), and to seed set (13%). Furthermore, partial coefficients of determination in the MANOVA also de-

HERRERA—BROOD SIZE IN THE GENISTEAE

893

Fig. 3 The most frequent pattern of seed production inside pods of Genisteae. Bars represent probabilities that ovules at varying distances from the style will set a seed, with lower position numbers corresponding to ovules closer to the style. Numerals by the bars are sample sizes (n = 100 ovules if no number is given). Horizontal dashed lines delimit 95% confidence intervals of mean seed set, and asterisks mark positions in which the probability of yielding a seed is beyond confidence limits for the corresponding sample size. For clarity, only the confidence interval for the maximum value of n is shown.

creased from ovules to seed set, indicating that the model’s ability to explain observed variation solely on the basis of species and population identity decreased from ovules to seeds and seed set. It might seem paradoxical that, given the increase in variance from ovules to seeds, only two species showed significant population differences regarding seed set, while eight did regarding ovule number. This is likely to result from a side effect of the relatively more stochastic nature of seed number and seed set compared with ovule number, namely decrease in statistical power (i.e., detecting a significant difference among two means requires a larger sample size). Thus, seed : ovule ratios often differed considerably among conspecific populations, but individual plants at each locality also varied considerably, and population means tended to remain statistically indistinguishable. Obviously, larger plant sample sizes may be needed to detect populational differences in this plant group.

Variations among Species Components of fecundity studied here were highly species specific, as demonstrated by taxon identity accounting for

nearly 90% of the observed variance in the MANOVA. However, there existed distinct modal values for ovule number, seeds per fruit, and seed set that shape a group-specific pattern. Ovule number might be expected to show moderate variations among species and genera because of a common phylogeny, but it was to some extent surprising that the average proportion of ovules that set a seed was so often in the range 40%–60%. I hypothesize that this results from largely stereotyped reproductive patterns within the lineage. Some or many species of this group probably have some kind of late-acting selfincompatibility (Rodriguez Rian˜o 1997), and one might expect them to have a mixed mating system. This would agree with both the extensive abortion of whole young fruits and also with extensive abortion of embryos inside developing pods. While accurate estimates of pod abortion are unavailable, data in this study indicate that within-pod abortion of embryos is often ca. 50% in the Genisteae. In association with low pollination success, much lower seed sets have been reported on occasion (e.g., Cytisus scoparius, Parker 1997; Cytisus multiflorus, Rodriguez Rian˜o 1997), but this is not likely to be

894

INTERNATIONAL JOURNAL OF PLANT SCIENCES

Fig. 4 Unusual filling profiles of Genisteae pods. See fig. 3 for captions

the case in the natural populations of this study in which bee pollinators are very common. With adequate pollinator service, even naturalized populations seem to observe the 40%–60% rule (50% for Chamaecytisus palmensis in New Zealand; Webb and Shand 1985). Exceptional in this regard are the low seed : ovule ratios of taxa with single seeded pods (e.g., Genista subgenus Voglera and species of Retama), and the unusually high values of Argyrolobium, Spartium, and Lupinus. Selective pressures related to seed dispersal and/or seedling establishment may result in the abortion of all embryos except one in fruits (Casper and Wiens 1981) and thus be responsible for the first anomaly. At the other extreme, Argyrolobium, Spartium, and Lupinus have been considered to be anomalous Genisteae by taxonomists on morphological grounds (Bisby 1981), and it is possible, therefore, that they might differ from the other genera also in their breeding system. Specifically, the results for annual Lupinus are in sharp contrast with those of most other Genisteae. Values near 90% found in this study support the notion that these taxa are largely self-compatible and have either autogamous or insect-induced self-pollination (e.g., Lupinus palaestinus and Lupinus pilosus, Pazy 1984; Lupinus nanus, Karoly 1992; Lupinus bicolor, Karoly 1994; Lupinus texensis, Helenurm and Schall 1996).

Position Effects within Pods Basal ovules set seed less frequently than distal (stylar) ovules in the majority of studied Genisteae. Since stigmatic pollen loads examined in a subset of species indicated that many more pollen grains germinated than were necessary to fertilize all ovules, this dominant, distal-skewed filling pattern is unlikely to result from insufficient pollination. Rather, it probably reflects higher embryo abortion rates at positions distant from the style, a frequent situation in the Leguminosae (Nakamura 1988; Bawa and Buckley 1989) and, in general, in plants with linearly arranged ovules (Bawa and Webb 1984). While Lee and Bazzaz (1986) related preferential abortion of proximal (basal) embryos to resource limitations within the fruit, this has been interpreted adaptatively by Bawa and Buckley (1989) as a way to increase progeny quality through differential abortion of less vigorous embryos following a phase of intense pollen tube competition within the style. Bawa and Buckley argued that basal ovules are likely to be fathered by slow-growing microgametophytes, which in turn would result in less vigorous (and thus more prone to being aborted) progeny being produced at the base. There is supporting evidence for this phenomenon in Phaseolus coccineus (Rocha and Stephenson 1991a, 1991b). According to the selective hypothesis, however, basal abortion should not occur in predominantly

HERRERA—BROOD SIZE IN THE GENISTEAE

895

Table 5 Average Numbers of Pollen Grains Recorded on the Stigmas (Attached to Fruits) of Five Genisteae Shrubs % Grains on the stigma Species

¯ X

s.e.m.

n

Grains/ovule

Grains/seed

Genista falcata .................. Genista polyanthos .. ... ... ... .. Retama monosperma ........... Stauracanthus genistoides .. .... Ulex eriocladus ..... ... ... ... ...

40 30 18 17 27

6 4 4 1 3

45 87 24 79 48

1.8 4.5 3.0 3.6 3.8

4.2 14.3 18.0 7.1 15.9

Note. See table 3 for fruit characteristics in these taxa.

autogamous species (fig. 3 in Bawa and Buckley 1989), while in the Genisteae it occurs in both outcrossers and selfers. Thus, and at least in this group, basal abortion might not be adaptive but be determined by vascular anatomy and represent simply a morphogenetic constraint (Watson and Casper 1984). Genetically or developmentally determined abortion patterns have been reported by a number of previous studies in other plant groups (Casper and Wiens 1981; Guth and Weller 1986; Gorchov 1988; Andersson 1990; Herrera 1990). Seed-filling patterns have been determined in this study by contrasting observed frequencies against confidence intervals, a method that, to my knowledge, had not been used previously. Most researchers have either regressed observed frequencies on ovule positions or used contingency tables to detect the region of the ovary where abortions occur, but this does not allow unambiguous identification of abnormally low- or highprobability positions inside pods (to say nothing of mere visual examination). If ovular positions are not tested individually, supposed patterns may lack materiality and hypotheses that

attempt to explain the reasons behind it can be meaningless. Bawa and Buckley (1989) recognize that the way a pod is filled with seeds will depend on the order of fertilization inside the ovary, so that, if it does not begin at ovules near the tip but at middle or basal ones, preferential seed abortion might respectively occur at both ends or at the tip. In this study, a few Lupinus species span a variety of filling patterns, indicating that research addressed to determine the order of fertilization in species of this genus may be helpful to understand how and if microgametophytic competition occurs in the Genisteae.

Acknowledgments R. Soriguer, S. Talavera, and E. Triano helped to locate and collect fruits from several Genisteae. Thanks are also due to P. E. Gibbs and an anonymous reviewer for comments on the manuscript. This study was supported by funds provided by Plan Andaluz de Investigacio´n, Junta de Andalucı´a, Spain.

Literature Cited Andersson S 1990 No evidence for selective seed maturation in Anchusa officinalis (Boraginaceae). Oikos 57:88–93. Arroyo J 1990 Spatial variation of flowering phenology in the Mediterranean shrublands of southern Spain. Israel J Bot 39:249–262. Arroyo J, T Maran˜o´n 1990 Community ecology and distributional spectra of Mediterranean shrublands and heathlands in Southern Spain. J Biogeogr 17:163–176. Bawa KS, DP Buckley 1989 Seed : ovule ratios, selective seed abortion, and mating systems in Leguminosae. Monogr Syst Bot 29: 243–262. Bawa KS, CJ Webb 1984 Flower, fruit, and seed abortion in tropical forest trees: implications for the evolution of paternal and maternal patterns. Am J Bot 71:736–751. Bisby FA 1981 Tribe 32: Genisteae (Adans.) Benth. (1865). Pages 409–425 in M Polhill and PH Raven, eds. Advances in legume systematics. Pt 1. Royal Botanic Gardens, Kew. Bossard CC 1991 The role of habitat disturbance, seed predation and ant dispersal on establishment of the exotic shrub Cytisus scoparius in California. Am Midl Nat 126:1–13. Buide ML, JM Sanchez, J Guitian 1998 Ecological characteristics of the flora of the Northwest Iberian Peninsula. Plant Ecol 135:1–8. Burd M 1994 Bateman principle and plant reproduction: the role of pollen limitation in fruit and seed set. Bot Rev 60:83–139. ——— 1995 Ovule packaging in stochastic pollination and fertilization environments. Evolution 49:100–109. Casper BB, D Wiens 1981 Fixed rates of random ovule abortion in

Cryptantha flava (Boraginaceae) and its possible relation to seed dispersal. Ecology 62:866–869. Eriksson O, B Bremer 1991 Fruit characteristics, life forms, and species richness in the plant family Rubiaceae. Am Nat 138:751–761. ——— 1992 Pollination systems, dispersal modes, life forms, and diversification rates in angiosperm families. Evolution 46:258–266. Fischer KE, CA Chapman 1993 Frugivores and fruit syndromes: differences in patterns at the genus and species level. Oikos 66: 472–482. Gorchov DL 1988 Effects of pollen and resources on seed number and other fitness components in Amelanchier arborea (Rosaceae, Maloideae). Am J Bot 75:1275–1285. Guth CJ, SG Weller 1986 Pollination, fertilization and ovule abortion in Oxalis magnifica. Am J Bot 73:246–253. Helenurm K, BA Schaal 1997 Genetic load, nutrient limitation, and seed production in Lupinus texensis (Fabaceae). Am J Bot 83: 1585–1595. Herrera CM 1989 Seed dispersal by animals: a role in angiosperm diversification? Am Nat 133:309–322. ——— 1990 Brood size reduction in Lavandula latifolia (Labiatae): a test of alternative hypotheses. Evol Trends Plants 4:99–105. ——— 1992 Historical effects and sorting processes as explanations for contemporary ecological patterns: character syndromes in Mediterranean woody plants. Am Nat 140:421–446. ——— 1995 Floral traits and plant adaptation to insect pollinators: a devil’s advocate approach. Pages 65–88 in DG Lloyd and SCH

896

INTERNATIONAL JOURNAL OF PLANT SCIENCES

Barrett, eds. Floral biology: studies on floral evolution in animalpollinated plants. Chapman & Hall, New York. Herrera J 1985 Nectar secretion patterns in southern Spanish Mediterranean scrublands. Is J Bot 34:47–58. ——— 1986 Flowering and fruiting phenology in the coastal shrublands of Don˜ana south Spain. Vegetatio 68:91–98. ——— 1987 Flower and fruit biology in southern Spanish mediterranean shrublands. Ann Mo Bot Gard 74:69–78. Jordano P 1995 Angiosperm fleshy fruits and seed dispersers: a comparative analysis of adaptation and constraints in plant-animal interactions. Am Nat 145:163–191. Karoly K 1992 Pollinator limitation in the facultatively autogamous annual, Lupinus nanus (Leguminosae). Am J Bot 79:49–56. ——— 1994 Inbreeding effects on mating system traits for two species of Lupinus (Leguminosae). Am J Bot 81:1538–1544. Kochmer JP, SN Handel 1986 Constraints and competition in the evolution of flowering phenology. Ecol Monogr 56:303–325. Lee TD 1988 Patterns of fruit and seed production. Pages 179–202 in J Lovett-Doust and L Lovett-Doust, eds. Plant reproductive ecology. Oxford University Press, New York. Lee TD, FA Bazzaz 1986 Maternal regulation of fecundity: nonrandom ovule abortion in Cassia fasciculata Michx. Oecologia 68: 459–465. Lloyd DG 1980 Sexual strategies in plants. I. An hypothesis of serial adjustment of maternal investment during one reproductive session. New Phytol 86:69–79. Lord J, M Westoby, M Leishman 1995 Seed size and phylogeny in six temperate floras: constraints, niche conservatism, and adaptation. Am Nat 146:349–364. Nakamura RR 1988 Seed abortion and seed size variation within fruits of Phaseolus vulgaris: pollen donor and resource limitation. Am J Bot 75:1003–1010. Parker IM 1997 Pollinator limitation of Cytisus scoparius (Scotch broom), an invasive exotic shrub. Ecology 78:1457–1470. Pazy B 1984 Insect induced self-pollination. Plant Syst Evol 144: 315–320. Pijl L van der 1972 Principles of dispersal in higher plants. 2d ed. Springer, Berlin.

Polhill RM, PH Raven, CH Stirton 1981 Evolution and systematics of the Leguminosae. Pages 1–26 in RM Polhill and PH Raven, eds. Advances in legume systematics. Pt 1. Royal Botanic Gardens, Kew. Primack RB 1987 Relationships among flowers, fruits and seeds. Annu Rev Ecol Syst 18:409–430. Renner SS, RE Ricklefs 1995 Dioecy and its correlates in the flowering plants. Am J Bot 82:596–606. Rocha OJ, AG Stephenson 1991a Order of fertilization within the ovary in Phaseolus coccineus L. (Leguminosae). Sex Plant Reprod 4:126–131. ——— 1991b Effects of nonrandom seed abortion on progeny performance in Phaseolus coccineus L. Evolution 45:1198–1208. Rodriguez Rian˜o T 1997 Biologı´a floral y reproductiva en Fabaceae de Extremadura. PhD diss. Universidad de Extremadura, Badajoz. SAS Institute 1990 SAS/STAT user’s guide, version 6, 4th ed. SAS Institute, Cary, N.C. Smith-Ramirez C, JJ Armesto, J Figueroa 1998 Flowering, fruiting and seed germination in Chilean rain forest myrtaceae: ecological and phylogenetic constraints. Plant Ecol 136:119–131. Stephenson AG 1981 Flower and fruit abortion: proximate causes and ultimate functions. Annu Rev Ecol Syst 12:253–279. Stephenson AG, RI Bertin 1983 Male competition, female choice, and sexual selection in plants. Pages 109–149 in L Real, ed. Pollination biology. Academic Press, Orlando. Talavera S, ed. 1999 Flora Iberica. Vol 7. Fabaceae, 1. Real Jardin Botanico de Madrid, CSIC, Madrid (in press). Tutin TG, VH Heywood, NA Burges, DM Moore, DH Valentine, SM Walters, DA Webb, eds. 1980 Flora Europaea. Vol 2. Cambridge University Press, Cambridge. Waser NM 1998 Pollination, angiosperm speciation, and the nature of species boundaries. Oikos 82:198–201. Waser NM, L Chittka, MV Price, N Williams 1996 Generalization in pollination systems, and why it matters. Ecology 77:1043–1060. Watson MA, BB Casper 1984 Morphogenetic constraints on patterns of carbon distribution in plants. Annu Rev Ecol Syst 15:233–258. Webb CJ, JE Shand 1985 Reproductive biology of tree Lucerne (Chamaecytisus palmensis, Leguminosae). N Z J Bot 23:597–606.